Quantification of disease progression and dropout for Alzheimer’s disease

Abstract

This research aimed to quantitatively describe the natural progression of Alzheimer's disease (AD) based on ADAScog scores in patients with mild-to-moderate AD. ADAS-cog data from 10 placebo-controlled clinical trials including more than 2,400 patients with up to 72 weeks of treatment were used. Different models describing the time course of ADAS-cog were evaluated. Patient characteristics potentially affect score changes were assessed. Furthermore, patient dropout patterns were characterized using parametric survival models. Covariate selection was performed to identify the risk factors associated with a higher dropout rate. ADAS-cog time course in mild-tomoderate AD patients receiving placebo was best described by a log-linear model, where the intercept represents the log-transformed ADAS-cog score at Week 10, the slope is the disease progression (i.e., natural increase of ADAS-cog score) on the log scale. Covariates influencing the intercept were baseline ADAS-cog score and baseline Mini Mental State Exam score. No covariates influenced the disease progression slope. A parametric log-normal model fit the dropout data best. Baseline ADAS-cog score and age were found to be significant predictors for dropout. AD disease and dropout models were both established. These models set up a quantitative basis for future clinical trial design and endpoint selection.